ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
(Generate patch)

Comparing Coro/Changes (file contents):
Revision 1.543 by root, Fri Oct 29 15:25:07 2010 UTC vs.
Revision 1.588 by root, Fri Nov 11 20:22:08 2011 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio (perl leaks like hell). 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: unready_all
5TODO: myhttpd header parsing
6TODO: channel->maxsize(newsize)? 4TODO: channel->maxsize(newsize)?
7TODO: http://www.microsoft.com/msj/archive/s2ce.aspx 5TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
8 6
76.07 Fri Nov 11 21:21:48 CET 2011
8 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
9 - update ecb.h.
10
116.06 Mon Aug 8 23:59:48 CEST 2011
12 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
13 backend, which is an order of magnitude slower. unfortunately, cygwins
14 pthread implementation isn't very complete either, so allocate the stack
15 twice just to be sure.
16 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
17 in its ithreaded perl - best recompile cygwin and use the 'w'indows
18 backend for much better performance. also disable ithreads for
19 even better performance...).
20
216.05 Thu Aug 4 21:36:36 CEST 2011
22 - blush, condvar values would not be propagated from send to recv anymore
23 (reported by Chip Salzenberg).
24 - use exponential increase for the readline buffer length in
25 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
26
276.04 Wed Aug 3 17:15:45 CEST 2011
28 - use even more efficient and more compatible condvars for
29 compatibility to AnyEvent 6.x :)
30 - more inconsequential ecb.h updates.
31
326.03 Wed Aug 3 11:41:30 CEST 2011
33 - change how Coro patches AnyEvent condvars for compatibility to
34 AnyEvent 6.x.
35 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
36 hasn't been included.
37
386.02 Wed Jul 13 04:35:19 CEST 2011
39 - "improve portability to Gentoo" - gentoo manages to put perl variables
40 in memory areas that are farther than 2gb apart, which the jit couldn't
41 handle and barfed. now it's just a bit slower on gentoo and similar
42 systems.
43
446.01 Sun Jul 3 12:31:14 CEST 2011
45 - workarounds are good, but the test for whether pthreads are used
46 was not good. thisone should be better.
47 - check differently whether gcc generates cfi instructions itself.
48
496.0 Wed Jun 29 19:43:35 CEST 2011
50 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
51 destroyed and cleaned up automatically (e.g. async { schedule }).
52 - implement a JIT compiler for part of the thread switch code,
53 which gives a 50% speed improvement on threaded perls, and
54 about 4% on non-thraeded perls (so threaded perls now finally
55 reach about half the speed of non-threaded perls).
56 - slightly modernise Coro::Intro, add section about rouse functions.
57 - avoid DEFSV and ERRSV, giving another 10% improvement
58 in thread switching.
59 - Coro::State->is_destroyed is now called is_zombie.
60 - implement a Coro->safe_cancel method that might fail, but
61 cancels in a "safer" way if it succeeds.
62 - add preliminary support for DEBUGGING perls.
63 - get rid of two hash-accesses when initialising a new Coro - this
64 speeds up coro creation by almost a factor of two.
65 - croak when a coro that is being cancelled tries to block
66 (e.g. while executing a guard block), instead of crashing or
67 deadlocking.
68 - use a more robust and also faster method to identify Coro::State
69 objects - speeds up everything a bit.
70 - implement Coro->cancel in XS for a 20% speed improvement, and to
71 be able to implement mutual cancellation.
72 - speed up context switches by a percent or two by more efficiently
73 allocating context stack entries.
74 - implement Coro->join and Coro->on_destroy in XS for a speedup and
75 a reduction in memory use.
76 - cancelling a coro while it itself is cancelling another coro is
77 now supported and working, instead of triggering an assertion.
78 - be a bit more crash-resistant when calling (buggy) on_destroy
79 callbacks (best effort).
80 - move on_destroy into the slf_frame, to allow extension slf
81 functions to have destructors.
82 - get rid if coro refcounting - simply crash in other interpreter
83 threads by nulling the pointers on clone.
84 - simplify warn/die hook handling when loading Coro - the convoluted
85 logic seems to be no longer neccessary.
86 - use libecb instead of our own home-grown gcc hacks.
87 - document alternatives to Coro::LWP. Please use them :)
88 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
89 sigaltstack. Really. wine suffers from it, erlang suffers from it,
90 and it's known since at least 2006.
91
925.372 Wed Feb 23 06:14:30 CET 2011
93 - apparently mingw doesn't provide a working gettimeofday, try to
94 work around that by relying on Time::HiRes (indirectly brought to
95 my attention by Max Maischein).
96 - fix some portability issues when Time::HiRes was used.
97
985.371 Mon Feb 21 14:36:08 CET 2011
99 - backport to windows process emulation code again.
100
1015.37 Sat Feb 19 07:49:44 CET 2011
102 - add a big "Coro thread life cycle" section to "man Coro".
103 - try a tentative workaround against the breakage that 5.13 has
104 introduced without depreciation period. sigh.
105 - no longer use Time::HiRes if gettimeofday is available, which
106 saves quite a lot of memory.
107
1085.36 Sun Feb 13 05:33:41 CET 2011
109 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
110 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
111 method call.
112 - undocument Coro::Timer::sleep and obsolete whole module.
113 - optimise Coro::Timer::timeout memory and cpu usage.
114 - slightly updated Coro::Intro for recent changes.
115 - do not initialise PL_dirty anymore.
116
1175.25 Thu Nov 11 01:08:39 CET 2010
9 - try a different approach on netbsd - netbsd 5 finally has marginally 118 - try a different approach on netbsd - netbsd 5 finally has marginally
10 working pthreads, but still broken ucontext/sigaltstack. 119 working pthreads, but still broken ucontext/sigaltstack.
120 - openbsd 4.8 finally got their act together, Coro works out of the box
121 with asm, setjmp and pthreads (no change, just informational).
11 122
125.24 Sat Oct 23 11:27:12 CEST 2010 1235.24 Sat Oct 23 11:27:12 CEST 2010
13 - port to the EV 4.0 API. 124 - port to the EV 4.0 API.
14 - work around bugs in mingw32, making strawberry perl work 125 - work around bugs in mingw32, making strawberry perl work
15 out of the box. 126 out of the box.

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines